A Comparison of Historical and Paleoseismicity in a Newly Formed Fault Zone and a Mature Fault Zone, North Canterbury, New Zeala

JOURNALOF GEOPHYSICALRESEARCH, VOL. 101,NO. B3, PAGES6021-6036, MARCH 10,1996

Acomparison of historicaland paleoseismicity in a newlyformed faultzone and a maturefault zone,North Canterbury, New Zealand

HughCowan 1and Andrew Nicol 2 Departmentof Geology, University of Canterbury,Christchurch, New Zealand

PhillipTonkin Departmentof Soil Science, Lincoln University, Lincoln, New Zealand

Abstract.The timingof largeHolocene prehistoric earthquakes is determined by dated surfaceruptures and landslides at theedge of theAustralia-Pacific plate boundary zone in NorthCanterb•, New Zealand.Collectively, these data indicate two large(M > 7) earthquakesduring the last circa2500 years,within a newly formedzone of hybridstrike- slipand thrust faulting herein described as the Porter'sPass-to-Amberley Fault Zone (PPAFZ).Two earlierevents during the Holocene are also recognized, but thedata prior to 2500years are presumedto be incomplete.A returnperiod of 1300-2000years between largeearthquakes in the PPAFZ is consistentwith a late Ho!oceneslip rate of 3-4 mm/yr if eachdisplacement is in therange 4-8 m. Historicalseismicity in thePPAFZ is characterized byfrequent small andmoderate magnitude earthquakes and a seismicityrate that is identicalto a region surroundingthe stmcturallymature Hope fault of the Marlborough FaultSystem farther north.This is despitean order-of-magnitudedifference in slip rate betweenthe respectivefault zonesand considerable dffœerences in the recurrencerate of largeearthquakes. The magnitude-frequencydistribution in theHope fault regionis in accordwith the characteristicearthquake model, whereas the rateof largeearthquakes in thePPAFZ is approximated(but overpredicted) by the Gutenberg-Richtermodel. The comparisonof thesetwo œaultzones demonstrates the importance of the structuralmaturity ofthe fault zone in relationto seismicityrates inferred from recent,historical, and palcoseismicdata.

Introduction earthquakesin thePorter's Pass-to-Amberley Fault Zone (PPAFZ), a regionof newly formed strike-slipfaulting that definesthe Earthquakehazard assessment traditionally incorporates the deformationfront at the edge of the Australia-Pacificplate analysisof historicand contemporaryseismicity, supplemented boundary zone in northeasternSouth Island. wherepossible by geologicalinformation about average slip rates, North Canterburystraddles the edge of the Australia-Pacific rupturelengths and slip per event. The timingand magnitude of plate boundaryzone betweenthe HikurangiTrough and the prehistoricearthquakes resulting in surfacerupture are determined Southern Alps (inset,Figure l a). Dextral strike-slipfaults of the bydating features offset by the fault and by examining the geometry MarlboroughFault System (MFS), farthernorth, transfer oblique ofthe fault trace.However, these data alone often do notprovide platemotion between the Alpine fault and the Hikurangi subduction sufficientinformation to characterizelarge palcoseismicevents zone. These faults have smaller cumulative offsets toward the [e.g.Ward, 1994]. southeastbut higher slip rates that reflecta southwardrelative InNorth Canterbury, New Zealand, we have supplemented data migrationof the locusof deformationacross the regionduring the fromfault traces with datafrom landslidesdistributed throughout late Quaternary[e.g., Yeats and Berryman, 1987; H. Andersonet al., theeastern foothills of the SouthernAlps. These data are used 1993].The mostsoutherly element of the MFS is the Hopefault collective!yto assessthe late Holocenechronology of large (Figurel a). Sixty kilometerssouth of the Hope fault, a zoneof disseminatedshear is evolvingalong a similartrend (Figure la) --l NowatNorwegian Seismic Array, Kjeller, Norway. [Rynn and Scholz, 1978; Carter and Carter, !982; t!emer and 2Now at Fault Analysis Group, Departmeat ofEarth Sciences, Univer- Bradshaw, 1985]. Elements of thiszone comprise anastomosing sityof Liverpool, Liverpool, England. strike-slipand thrustfaults that extend east-northeastfrom the SouthernAlps along the northern margin of theCanterbury Plains Copyright! 996 by the American Geophysical Union. andpass into a fold andthrust belt nearthe townof AmberIcy PaperNumber 95JB01588. (Figuresla and2). We usethe name Porter's Pass-to-Amber!ey 0148-0227/96/95JB.0!588$05.00 Fault Zone (PPAFZ)to describethe strike and approximate

6021 6022 COWANET AL.: PALEOSEISMICITYIN A NEWLYFORMED FAULT ZONE

! i !

• I I /./"

1170E Figure la. Major structuralelements of the northeasternSouth Island, andthe locationsand years of selectedhistor- ical earthquakes.The onshorestructure is modifiedfrom Gregg[1964], Cowan [1992] and unpublisheddata of the Active TectonicsResearch Group, University of Canterbury.Offshore structure is simplifiedfrom Barnes [1994]. The numberedrectangles delineate two regionsselected for analysisof historicalseismicity. Abbreviations are MFS, MarlboroughFault System;HPF, Hope fault;HRS, HopeRiver segment;KPF, Kakapo fault; PPAFZ, Porter'sPass- to-AmberleyFault Zone;PBF, PegasusBay fault;PP, Porter's Pass; and AY, AmberIcy.Inset showslocation of Fig- ures la and lb in relationto the Alpine Fault (Air) and Hikurangi Trough (HT), the main elementsof the New Zealandplate boundary.The arrowindicates the directionand magnitude(42 mm/yr) of the plate motaonvector [DeMets et al., 1990]. geographicend-points of the zone,which has only recently been HistoricSeismicity of the North CanterburyRegion mappedcomprehensively [Cowan, 1992]. The region betweenthe Hope fault and PPAFZ is one of Earthquakeswithin and adjoiningthe North Canterburyregion transitionfrom subduction to collision[Reyners and Cowan, 1993], duringthe period1942-1994 are shownin Figurelb, withthose wherethe geological structure of theupper crust is dominatedby greaterthan magnitude 6.5 since1888 indicated on Figure la. The northeaststriking reverse and thrust faults that dip southeast and are focaldepths for most events in thecatalogue are uncertain due to the closelyassociated with growingfolds [Yousif, 1987; Nicol et al., widespacing of nationalnetwork seismographs (typically >50 kin), 1994]. This region has accommodated~12-15% northwest- butlocal studies of microseismicity,recorded on portable, small- southeastshortening during the Pleistocene, and rates of shorteningaperture seismograph arrays, have indicated that most earthquakes doseto thePacific coast are about 1%/100 kyr [Nicolet al., 1994]. inthis region are restricted to theupper 10-15 km of thecrust [Rynn Detailedstudies along the edge of theplate boundary zone in North andScholz, 1978; Cowan, 1992; Reyners and Cowan, 1993]. Canterburyindicate that deformation probably commenced within TheHope River segment of theHope fault (Figure la) andan thelast 0.5-1 m.y., disrupting a volume of crustnot significantly adjacent splay (Kakapo fault) have each ruptured during historic deformedsince the Cretaceous [Cowan, 1992; Nicol et al., 1994]. earthquakesof magnitudeM---7.3 and M~7.0, respectively [Eiby, Thepurpose of thispaper is to evaluatethe late Holocene record 1968;Cowan, 1991; Yang, 1991, 1992].Two moreevents of of large earthquakesin the PPAFZ, in order to understandthe magnitude6.5.-6.9 have occurred in the easternpart of North frequency-magnitudedistribution ofearthquakes in anewly formed Canterbury,in 1901 and 1922, and another event of M•, 6.7 occurred faultzone. We comparehistorical and palcoseismic data from the in June 1994, to the west of the PPAFZ. The 1994 earthquake PPAFZwith data from the adjacent structurally mature Hope fault occurredin the upper crust, and the 1901 and 1922 eventsare zone, and use thesedata to evaluate differencesin fault behavior presumedtohave also been shallow, based on relatively limited felt betweenstmcturally mature and newly formed fault zones. areas[McKay, 1902; Skey, 1925]. Many more faults in North COWANET AL.-PALEOSEISMI•• IN A NEWLYFORMED FAULT ZONE 6023

O0 o

Figurelb. ShaLlowseismicity (<40 km) for theregion of Figurela duringthe period 1942-1994 (May 30) [New ZealandSeismological Observatory, 1994]. Symbols are scaled with magnitude, and the catalogue iscomplete above M 3.0since 1988; M 4.0since 1964; M 5.0since 1942. The dense clusters of seismicity,north of theHope fault and westof thePPAFZ, are associated with earthquakes of M w 5.9 andM w 6.7, in 1990and 1994, respectively [Ander- sonet al., 1993; ,•rnad6ttir et al., in press].

Canterburyexhibit evidence of Holocenedisplacement [Gregg, Oligoceneand lower Miocene timestone beds across the Mount 1964],and since 1964, the New Zealand national seismograph Grey block and vector analysis ofslip across Mount Oxford (Figure networkhas locatedfrequent earthquakes of small to moderate 2) (H.A. Cowanet al.,manuscript in preparation1995). magnitude(M<5.4) along the PPAFZ, making this one of thebetter The Porter'sPass fattit forms the principal western element of definedsurface fault zones in NewZealand [Reyners, 1989]. Data thePPAFZ and is associatedwith Holocene offsets near Porter's fromthe instrumental catalogue are utilized later in thispaper for 'Pass[Speight, 1938; Wellman, 1953; Berryman, 1979; Coyle, 1988; comparisonsof recurrence parmeters for largeearthquakes within Knuepfer,1992]. New dataon lateHolocene faulting from this area andbetween the Hope fault and PPAF'Z regions. are presentedhere. Farther east, the Porter's Passfault bifurcates aroundMount Oxford,and splays extend northeast and east to Lees Porter'sPass-to-Amberley Fault Zone Valleyand the northern Canterbury Plains, respectively (Figure 2). A surfacetrace in LeesValley is associatedwith scarpsup to 6 m ThePPAF-Z extends east-northeast from the Southern Alps and highon Holocenealluvial fans [Gregg, 1964; Garlick, 1992],and passesinto a fold and thrustbelt near the town of Amberleyand along the southernflank of the Ashley Range,major zonesof offshore(Figures la and2) [Barnes,1994; Nicol et al., 1994]. crushedrock diverge to the eastand southeast(Figure 2). The east Principalelements of thePPAFZ bound the northern margin of the strikingzone extends into a regiondominated by thrust faulting and CanterburyPlains where the foothills of theSouthern Alps rise to foldingcentred on MountGrey, whereas the southeast striking zone 2000m. ThePPAF'Z is definedby zonesof crushedTorlesse passes beneath the CanterburyPlains to the Cust anticlineand greywackethat strike east and northeast,at a tugh angleto the Ashleyfault and may extend offshore to thePegasus Bay fault zone regionalNW-SE grain of Mesozoic deformation [Gregg, 1964], and farthereast [Kirkaldy and Thomas,1963; Herzer and Bradshaw, locallycut through the late Cretaceous-Cenozoic coversequence, 1985; Barnes, 1994] (Figure la). Dataon lateHolocene faulting and disruptinglate Pleistocene and Holocene landforms. A total offset landslidesin theeastern part of thePPAFZ have been collected from of<2 km across the PPAFZ is inferred from the strike separation of Custanticline, Ashley fault, andMount Grey. 6024 COWANET AL.: PALEOSEIS•CITY IN A NEWLYFORMED FAULT ZONE

Lees Valley

'"" Ashleyfault '

I Oxford I t ... I I L Fig.4

FAULTS FOLDS LANDSLIDES inferred syncline• Dated Undated Canterbury Plains strike-slip •-- I © o.•--•o-•.•o%3 thrust • anticline• N surface trace xxxxxx © o-oo_1 1Okra I I ___•_ _ i.i:•urban aroa •...... BasementI I Coversediments highway ß 2195olovation (m) i Figure 2. Structuralelements of the Porter'sPass-to-Amberley Fault Zone (PPAFZ) in North Canterbury,New Zealand,showing the locationsof main faults,folds, and prehistoric landslides. Faults that are depictedwithout sense-of-movementindicators represent mapped zones of crushedrock. Numbered landslides correspond to datain Table2 andFigure 6. Thindashed lines denote the margins of activefolds on theCanterbury Plains. Abbreviations are KA, Kowai anticline;CA, Custanticline; MO, Mount Oxford,and MG, Mount Grey. The locationsof Figures3, 4, and 5 are indicated.

Holocene Faulting in the PPAFZ heightof 150 m on thenorthern flank of Custanticline. A steep north-to-southgradient of Bouguergravity anomalies across the Porter's Pass Fault anticlinehas beenattributed to the presenceof a faultwith significantthrow [Kirkaldy and Thomas, 1963]. We infer that this Studiesnear Porter's Pass,where the PPAFZ is relatively narrow, faultis connected to the southeast striking extension of the Porter's indicatea Holoceneslip rate of ,--4mm/yr [Berryman, 1979] and Passfault at GlentuiRiver, which appears to controlthe fl0w 2.7-3.8mm/yr [Knuepfer, 1992]. A few hundredmeters north of directionof the AshleyRiver in the areaimmediately to the Highway73 atPorter's Pass (Figure 2), a radiocarbondate from the northwestof Custanticline (Figure 4). lowerpart of a peatcolumn accumulated behind the surface trace LateQuaternary uplift of Custanticline is evidentfrom drainage indicatesthat the surfacetrace probably formed 2000-2500 years featurespreserved across the crest of thefold, which is capped by B.P. (Table 1). two late Pleistoceneloess sheets [Trangmar, 1987]. Ancestral In a highwayembankment on the southside of the pass,two channelsof the AshleyRiver are incisedinto the loesssheets buriedsofts, each containing charcoal, are offsetacross the fault (Figure4), sothe diversion of AshleyRiver to its present course has (Figure 3). The crosscuttingrelationships together with the presumablyaccompanied uplift at the westernmargin of the stratigraphyindicate possibly two rupturesduring the Holocene anticline,possibly during the earlyHolocene since the youngest (Table 1). The olderof the two soils,offset across fault 1, may ancestralchannel has no loesscover. The upliftwas probabl5 reflecta rupturebetween 7000 and 9000 years ago. Faults 2 and3 causedby rupture on a faultthat is associated with a 1.5-kin- long havedisplaced both soils at leastonce since 7000 years ago. surfacetrace (Figure 4) anda large(-500 m) offsetin basement Cust Anticline [Kirkaldyand Thomas,1963]. Probablelate Holocene ruptures in thisarea are inferred from TheCust anticline andAshley fault, located ona major southeast dated landslides andthe timing of Ashley River downcutting along splayof thePPAF-Z, areimportant sites for palcoseismic thenorthern flankof the anticline. Woodcollected fromthe base0f investigation(Figures 2,4). The recent nature ofthe deformation in gravel deposits overlying thestrath ofa terrace 13+ 12m above thisregion isindicated bythe presence ofconsolidated Pliocene- theAshley River was dated atbetween 2100and 2500 calibrated earlyQuaternary siltstone andconglomerate whichcrops out to a radiocarbonyearsold (Table 1and Figure 4), implying anaverage , COWANET AL.: PALEOS•-•SMI• •N A NEWLYFORMED FAULT ZONE 6025 6026 COWANET AL.: PALEOSEISMICITYIN A NEWLY FORMEDFAULT ZONF_,

ß.'.'.'.'.' ' Nothofaguscharcoa!'.'.' ß ...... 500-700 •,ea½sB.'P.'.'-'- '.: . '•..'.', ,'-'.'.'-'.'.', 16900:L--90 I ß '- ! ' ";' "';"'''";"'I(NZ 7053)1

ß's0il'.' .' ß ' ' i' i' ib•ri•d ''];: '

,.t-11o .

ß ...... SOil.'.'.' ß '.' .'.'.' ß.-.'.'.- ...... ß

......

...... -..

ß ...... '-'-'-.-'.IFAuLT2I--•.': -:-:-:-

Figure 3. Age relationsbetween faults and buried soils exposed in a road embankmentwithin the Porter'sPass fault zone at Porter'sPass. See Figure 2 andTable 1 for sitelocation and description, respectively.

downcuttingrate of 5.7 _+ 1.4 m/kyr. The heightof the terrace River is similar to the Ashley River, then the unfaultedterrace strathabove the presentfloodplain presumably reflects an element formed'-'600-900 years ago. The availabledata thus imply that the of tectonicuplift andfold growth,because upstream the fiver is in lastrupture of the Ashleyfault postdated 2500 yearsB.P. and could equilibrium,whereas downstream it is aggrading[Cowan, 1992], have accompaniedthe growthof Cust anticlineprior to 600-900 Whether or not the downcuttingby Ashley River reflects yearsB.P. continuousor episodicuplift is unclear,but additionaldata from landslidesat thislocality (Figure 4) suggestthat the first increment Mount Grey Fault of uplift was inducedby coseismicrupture in this sectorof the PPAFZ. TheMount Grey fault comprises a northdipping thrust along the southside of MountGrey (931 m) in theeastern part of thePPAFZ Ashley Fault (Figure2). The fault loopsaround the eastern flank of MountGrey andbifurcates, forming an oblique,left-lateral tear fault that strikes The Ashley fault offsetsPleistocene and Holocene terraces of the north and dips to the west, and againstwhich the Tertiarycover OkukuRiver eastof Custanticline and has a surfacetrace length of sequencehas been overturned [Wilson, 1963]. The fault is exposed 4.5 km [Brown, 1973; Bero'man, 1979] (Figure4). The fault is in a numberof streamsand forms a prominentsurface trace and downthrown to the south, with throws of 1-4 m on a late Pleistocene hillside bench hollow, behind which scree and bouldershave aggradationsurface and 0.5-1.5 m on a Holocenestrath terrace accumulated. [Cowan, 1992]. Differences in the throw of the fault on thesetwo Radiocarbonand weathering-find dates delineate two surface surfacessuggest at leasttwo rupturessince the formationof thelate ruptureevents within the last 2500 years (Table 1). Evidence for the Pleistocene aggradation surface. Variations in throw which oldestrecognized event was obtained from a streambank exposure accompanyslight changes in the strikeof the fault, are consistent of thefault (Figure 2, 5), wherewood and peat were preserved on with a minor fight-lateral componentof displacement,but no thesurface of a debrisflow, offsetacross a faultscarp that is now unequivocaloffsets have been documented. buried3 m belowthe present ground surface. The wood was not in The height of the Holocenestrath terrace offset by the Ashley growthposition and was buried beneath younger colluvium, which fault is similar to the dated (i.e., 13 m) strath aboveAshley River, in turnwas offset by the fault. A soil,developed on the debris fl0w so the terracesmay be of approximatelyequal age.The fault trace andcontaining a traceof charcoalof unknownage defined an acrossthe Okuku River strathis not significantlydegraded, so the apparentvertical offset (downthrown to the west) across the fault. fiver presumablyabandoned the surfaceprior to, or at the time of, Wecould not discern from the field relationships whether ornot this faulting.The oldestOkuku River terracenot offsetby thefault (its buriedsoil was deformed orhad merely developed onan exposed riser truncatesthe surface trace) is about 4 m above the active faultscarp at a later date. The soil was buried beneath another debris floodplain(Figure 4). If the averagedowncutting rate of the Okuku flow,also capped by a youngersoil profile containing a tree stump COWANET AL.: PALEOSEISMICITYIN A NEWLYFORMED FAULT ZONE 6027

KOWAI ANT/CLINE

-"-" / ..."*

...... -- , , • - -•

•% , _, -'_ I•//,1',;•%¾ l •½ ...... •"'•1 ,- / , ß __ •.• " k ', • •---- ) • ...... - •

I ,

,,eQuat. sediments '• Cret-earlyß Ouat ' sediment• -- __ • :'.:•Torlessegre•ackebasement Fold• •11 ) :• % ; ,, " •1 ' I •72-• '• .... Figure4. Map of thenorthern Canterbury Plains showing the relationships between the Cust anticline, Ashley fault, and range-boundingelements of the PPAFZ. LSa andLSb indicate locationsof the radiocarbondates from landslide 8 (Table 2 and Figure6). Topographiccontours are reproducedat 20-m and 10-m intervals(thin solidand thin dashedlines, respectively) from NZMS 270, sheetsL35 andM35, by permissionof the Departmentof Surveyand Land Information, New Zealand. ingrowth position. A radiocarbondate of 182 + 60 yearsB.E (0.2-10x 106m 3) rockavalanches, derived from outcrops of obtainedfrom an outer portion of thetree stump (Table 1) indicated Tofiessebasement (Figure 2). Thedata indicate a clusteringof ages a minimumage for the buffedsoil andthe last surface rupture betweenabout 500 and700 yearsB.P., with othersscattered back to recognizedat this site. morethan 7000 yearsB.P. (Table 2 andFigure 6). Chipswere collectedfrom Torlessesandstone boulders and Landslidesin the range 500-700 yearsB.P. are relatively cobblesexposed at theground surface along the fault trace north of uniformlydistributed along the PPAFZ and are well preserved. thestream bank locality, and weathering-rind ages for these were However,there are insufficientdata to confirma similarspatial calculatedusing the method of Chinn[1981] and calibration curve patternof landslidesat 2300 yearsor older,which presumably ofMcSaveney [1992] to evaluate the timing of site disturbance. Two reflectsthe removal of landslidedebris by erosion.A largelandslide modalrind thicknesses were obtained and the ages (Table 1) arein identifiedon the north limb of Cust anticline(Figure 4), which goodagreement with thetime intervals for which surface rupture is containedwhole trees and an articulatedskeleton of the largest inferredfrom the radiocarbon ages. The available data collectively species of extinct moa, Diornis giganteus[Cowan, 1992], was implysurface rupture on the Mount Grey fault, between 2300-2400 radiocarbon dated at circa2300 yearsB.P. (Table 2). Thisage is the yearsB.P. and 300450 yearsB.E sameas the nearby strath terrace of AshleyRiver (Figure 4), andwe surmisethat both the landslide and the onsetof fiver downcutting accompaniedcoseismic uplift of Cust anticline. Pre!fistoricLandslides of the PPAFZ To assesswhether the effects of thisevent extended beyond the geographiclimits of ourstudy, the ages of rockavalanches from the Landslidesprovide additional data with whichto assessthe central SouthernAlps [Whitehouseand Griffiths, 1983] were recurrenceinterval of largeprehistoric earthquakes in thePPAFZ. recalculatedusing the weathering-rindcalibration curve of Twentylandslides have been identified within the PPAb-Z, and the McSaveney[1992]. Two modal ageswere obtainedfor Southern agesfor 10 of thesehave beendetermined from radiocarbonand Alps landslides(circa 1050-1400 yearsB.P. and ckca 1700-2100 weathering-rinddating (Table 2). Thelandslides are mostly large yearsB.P.), and these do not coincide with modal ages for landslides COWANEl' AL.: PALEOSEISMI•• I:NA NEWLY FORMEDFAULT ZONE

/'-:.••...:...... i•ii•!!::::i::::::•:!ii!i::::•::::ii•!ii•::•!•i;!!!!i!!i::i:•::•'•.'-':-'-:'ßß- lDebris f•ow] ======

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::•BuriedsoilsJ ======

.. _ • • _ _ _ '• .....•:.:::.•:::::.• ...... :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

x '•.'.•1::::::•.:.,•.• -:.:.:.:.:.:-:.:.:.:-..:-.-'-:-'-'ß....:.:.:. [Debrisflow ] ,..,•• '•:-',:.:::s...... '"', '.... ' " ':':' ß I Co•uviuml ' '- :"-• .... ======-I J: ' / ...,•,..--,',',':•-.======::::::::::::::::::::::::::::::::::::::::::: "::::::i::':i:::'::-'.':-,-"•;:'•... ======...... :::...... :::::::::i::::::: :::::::::::::i:::!:::::::::::::::::::::::::::::::::•...... :...•:::::::::::.•:::::: ...... ::::::i::::::::..,:::: ::I:::::: ...... :::i::::::":.<:""':•..'...... :::•,.....-':••:•....:.....•:••...... •...... ,•:-,..--.'••":...... "•:••...... '•.....-::••:..,.:,.•_"'"'•'-"-'•'-' ,•'"'' ._...... :•....) ...... •.x::'::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::...... :...... ,:::...... :,: ::...... ,:. :::::::::::::::::::::::::::::::::...... ::::...... :: ...... :::: Crushed ...... :...... oreywacke} - ...... :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::•... :<•...... ,...... ::::::::::::::::::::::::::::_'-"...... <...... ':':':':': :':" T ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ';(:':'"':':':" :...... :::--•:...... W •...... "':'":'"":'"'""'"" ::...... ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::,.. :...... ::...... •,::...... :...... :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::...... -x..-.'.'.'-'.'...... '.'.'.'.v.'.'.:.•'.'-'.'.'..,i.'-'.'.'.'.'.'-'.'-'.<: ...... '"".:.:.:.:...•:.:.:..'""•••••••::i..•"'"'"'""'""" :.:.:.: :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ......

--• '" [Streamchanneli -• I Mt Grey Fault ] 2330+_40yrs B.P 0 1 (NZ 7852) . metres Figure5. Stratigraphyand conventional radiocarbon ages associated with samples from an exposure of theMount Greyfault in a streambank on theeastern flank of MountGrey. See Figure 2 andTable [ for sitelocation and description,respectively. documentedin thePPAFZ (Figure6). The agesfor threelandsfides periodof 3 monthsand were followed 39 yearslater by theM, 7.4, fromthe central Southern Alps do overlapwith thoseof thePPAFZ, Mw 7.1 Inangahuaearthquake [Adams et al., 1968;Anderson et al., in therange ~2200-2450 years B.P., but fault data indicate probable 1994]. It is doubtful whether each of these events could be ruptureof thePPAFZ during that period (Table 1). differentiatedfrom the geologicalrecord using radiocarbon dating Six landslideswith agesbetween 500 and 700 yearsB.P. are had they occurred during prehistoric time, although locatedwithin 1 km of the majorfaults of the PPAFZ,and two of dendrochronologyand lichenometrycould potentiallydiscriminate thesebury the traceof the Porter'sPass fault nearPorter's Pass the 1929 and 1968 events[e.g., Bull et at., 1994]. Second,as the (Figure2, landslides1 and2 ). By contrast,only two landslides elapsedtime sincethe creationof a landformincreases, so too does withinthis age range were dated by Whitehouse and Griffiths [1983] the likelihood that it will be destroyed,modified, or buriedby in the SouthernAlps farther west. Collectively,the spatial subsequentprocesses. Furthermore, periods of nondeposition may distributionof these data suggeststhat if the landslideswere concealthe occurrenceof ruptureevents at a given site,and seismicallytriggered, they are more likely related to local rupture of assumptionsof continuoussedimentation are difficult to testand are thePPAFZ rather than to a regionalevent on theAlpine Fault, an rarelyproven [e.g. Doig, 1990;Cowan and McGlone,1991]. We interpretationfavored by W.B. Bull (personalcommunication, cannot rule out periodsof non depositionat any of the sites 1994)based on recent1ichenometric dating of landslidesin the describedin thisstudy, but the temporal clustering of landslidesin SouthernAlps. therange 500-700 years B.P. and their close proximity to the PPAFZ Theinterpretation of landslide distributions in paleoseismology suggest that the most recent rupture of thiszone probably occurred involvestwo assumptions:(1) that landslideswere triggered duringthat period. The landslides attributed to earlierevents (2000- coseismically;and (2) that all landslidesof a similarage were 2500years B.P. and 7000-9000 years B.P.) are much less abundant, coseismicwith the sameevent. Landslides have beentriggered by indicatingthat in New Zealand'smountainous terrain, landslide all large(M >7) historicearthquakes in New Zealand,but many datamay only be usefulfor characterisingpaleoseismicity for up to largelandslides in seismicallyactive regions, including the 2000-3000yearsB.P. SouthernAlps, havenot beentriggered by earthquakes[e.g., Plafkerand Ericksen, 1978; McSaveney et al., 1992;McSaveney, Magnitude Estimates 1978,1993]. In theabsence of independentevidence, the coseismic interpretationof prehistoric landslides must therefore be regarded Thedating of surfacefault ruptures provides a check on the astentative [Whitehouse and Griffiths,1983]. likelihoodthat a givengroup of landslideswas triggered For both fault tracesand landslidesthere are at least two reasons coseismically,butthe estimation ofmagnitude isproblematic inthe whyevidence for large palcoearthquakes could be sparse. First, it absenceof datathat define the rupturedimensions, and these mustbe rememberedthat few datingtechniques can discriminate parametersare poorly constrained for thePPAb-Z. Relationships eventsseparated by a fewyears or decades,a point exemplified by betweenlandslide distributions and earthquake magnitudes have threehistoric earthquakes in northern South Island. Both the Ms 7.8 beenproposed based on global compilations ofdata from historic Bulletand M s 7.0 Arthur'sPass earthquakes of 1929 [Henderson, events [Keefer, 1984], but theseprobably do not providean 1937;Speight, 1933; Dowrick and Smith, 1990] occurred within a appropriatebasis for estimatingthe magnitude of prehistoric COWANET AL.: PALEOSEISMICITYIN A NEWLYFORMED FAULT ZONE 6029

Table2.Description ofDated Landslides From the Porter's Pass-Amberley FaultZone

LocalityFrom LandslideVolume, Figure2 Latitude,Longitude 106 m 3 Agea Methodb AgeCalendar Years ø

1, Achcron Rivera 43.31,171.66 6.0 500 +_69 e 14C A.D. 1290-1515f two slidese NZ 547

2, KowaiRiver • 43.28,171.78 2.0 560 + 90 WR A.D. 1340-1520 60, 1988

3, KowaiRiver• 43.28,171.80 0.2 590 + 100 WR A.D. 1300-1500 40, 1988 4, Coal Creek 43.40, 172.00 0.02 5600 + 62 •4C 4575-4335B.C. f NZ 7918

5, AshleyGorge 43.19, 172.14 0.3 490 + 70 WR A.D. 1430-1570 75, 1991

6, LeesValley 43.11,172.12 10.0 580 _+ 90 WR A.D. 1320-1500 62,1991

7, GlentuiRiver 43.21,172.26 4.0 600 +_ 100 WR A.D. 1290-1490 two slides 1210 + 190 70, 1991 A.D. 590-970

45, 1991 8a,Cust Anticline 43.26,172.37 0.1 7400+ 90 i4C 6425-6000B.C 'f NZ 7854 8b,Cust Anticline 43.26,172.37 3.0 2300+ 60 14C meanof two dates NZ 7855 200-400B.C. f 2270 + 82 14C NZ 7856 9, MountThomas 43.15,172.40 4.0 2467+ 66 I•*C 400-765B.C 'f NZ 7895

10, MountGrey 43.09,172.55 0.2 3430 _+ 470 WR 970-1400B.C. bimodal 99, 1991 1930-3250 B.C. 4580 +_ 660

SeeTable 1 footnotefor explanationof weathering-rinddating errors. aYears before A.D. 1950,with modalweathering rind agesrounded to nearest5 years. bSymbol •4C denotes conventional radiocarbon ageand sample numbers; WR is weathering rinds, followed by number ofrinds and year (A.D.) of measurement. 14 deSymbolData from CBurrows ages (,calendar[1975]. years A.D./B.C.). e Datafrom W.B. Bull (writtencommunication, 1994). f Calibratedradiocarbon ages (95% confidence levels unless otherwise stated) based on compilation by$tuiver and Reimer [I986] of 20-year tree ringdata for the period 7210 B.C. to A.D. 1950, with offset of-30 radiocarbon years as recommended by$tuiver and Pearson [ 1986] and Pearson and $tuiver[1986]. •Datafrom Coyle [1988]. earthquakes.Only the largestprehistoric rock falls may be Thethreshold magnitude for surfacerupture and coseismic recognized,andvariations in earthquake source mechanism and landslidingin the PPAPZ is unclear.However, since all historic regionalgeology could strongly influence the distribution of eamhquakeslarger than magnitude M •-7 in northernSouth Island landslides.For example,the 1929 and 1994 Arthur'sPass havebeen accompanied by landslides[McKay, 1902; Henderson, earthquakeseach triggered rock fails over an areacomparable to 1937;Speight, 1933; Adams et al., 1968] we infer that the late manyglobal events of similarmagnitude (M. McSaveney,personal Holocene landslides in the PPAFZ were probablytriggered by communication,1995), but the areas affected by large landslides of earthquakesof similar magnitude.More precise estimatesof the1929 Buller and 1968 Inangahua earthquakes [Adams, 1981] are magnitudewould depend critically on detailedknowledge of the smallerthan might be expected from Ke•er's global catalogue. The sub-surfacefault dimensionsand segmentation[Wells and Bu!lerand Inangahua earthquakes occurred in a regioncomposed Coppersmith, 19941, parameters that are poorlydefined in this largelyof crystalline Paleozoic rocks, in contrastto the weakly hybrid strike-slip and thrust fault zone. Indeed, lateral variations in mtamorphosedgreywacke of the SouthernAlps. These fault behaviorare to be expected,and historic earthquakes 'differences,together with variationsin focaldepth, rupture elsewhere [e.g., Arnaike, 1987: Stein and Yeats,1989; Hull, 1990; directivity,and topographiceffects illustrate some of the Steinand Ekstrem,!992; SrnaIleyet aI., !993; Andersonet aI., unities thatwould accompany the estimation of earthquake 1994; Jibson et al., 1994]suggest that notall Holoceneruptures in magnitudesfrom prehistoriclandslide distributions. The the PPAPZ would necessarilyproduce surface fault offsets uncertaintiesnaturally would be compoundedby incomplete indicative of earthquakemagnitude or amenableto palcoseismic aamp!ingofthe spatial distribution ofprehistoric landslides. investigation.This seems especially likely in thefold and thrust belt 6030 COWAN ET AL.- PALEOSEIS•CITY IN A NEWLY FORMED FAULT ZONE

...... •. pPAFZ"raPix,re-,•: N:34 • • 8-

• 0 o d

<• • •v •- CA+A

N • --MG

• • rag* -- PP • • • i•I I I I

' 'D m -- G

• I'2 •9 o 10 10 • • • • 7 ' • 8b ......

I I ...... • ß I

0 1000 2000 3000 4000 5000 Years B.P. Figure6. Temporal distribution oflandslides andfault ruptures inthe PPAFZ, and a comparisonwithdated land- slidesfrom the Southern Alps[Whitehouse andGrifflths, 1983]. Weathering-rind agesfrom the Southern Alpshave beenrecalibrated usingthe calibration curve ofMcSaveney [1992]. Landslide numbers correspond tothose inTable 2 andare located onFigure 2.Abbreviations areA, Ashley fault; CA, Cust anticline; MG, Mount Grey fault; PP, Porter'sPass fault. Asterisks denote fault ruptures dated by weathering rinds. Refer to Tables 1 and 2 fordetails of faultruptures and landslides, respectively. at the easternend of the PPAFZ and farthernorth, where late log N(M) = a - am (1) Holocenesurface traces are preserved, but locally display vertical offsetsthat are anomalously large with respect to faultlength and where N(M) is the cumulativenumber of earthquakesof for whichno landslideshave been correlated [Cowan, 1994; A. magnitudeequal to or greater than (M); a definesthe rateof Nicol et al., unpublisheddata, 1991]. Uncertaintiesin the occurrence,and b is the exponential decay with increasing assessmentof fault dimensionsand the likelihood of hiddenfaults magnitude,implying self-similarity [e.g., King, 1983]. The inthis young fault zone also complicate theestimation of slip rate parametersa andb are constantsof the seismicitymodel and are andpotential slip per event. Local variations in faultkinematics are computedfrom a catalogueof historicalseismicity. Equation (1) evidentfrom the differential uplift at Cust anticline (this study) and maybe integratedto obtaina cumulativeexponential distribution, in thefold and thrust belt farther east [NicoI et aI., 1994].Davis and truncatedat a maximummagnitude for a givenset of faultsor Namson[1994] and Shaw and Suppe [1994] have shown that volume of crust, suchthat deta•edstructural analysis can reveal the presence ofhidden faults, N= No[10a(Mø' •) - 10b(Mø' Mmax)] (2) andsimilar work in progressin thePPAFZ (H.A. Cowanet al., manuscriptin preparation, 1995) may provide further insight into NO is thenumber of earthquakesof magnitude Mo orgreater, thekinematics and subsurface structure of the zone, especially if whereM o is an arbitrarythreshold magnitude [e.g., Youngs and combinedwith recent observations ofgeodetic strain (C.F. Pearson Coppersmith,1985]. When combinedwith an appropdm et al., Straindistribution across the Australian-Pacificplate attenuationexpression, the seismicity model may be used to derive boundaryinthe central South Island, New Zealand, from 1992 GPS returnperiods orprobabilities ofexceedance forspecified levelsof andearlier terrestrial observations, submitted to Journal of groundmotion. This classical hazard analysis, advanced byCornell GeophysicalResearch; hereinafter Pearson et al.; submitted[1968], has been applied widely in seismic hazard assessment [• manuscript,1995). Here, we restrict our discussionto the J.G.Anderson etaI., 1993]and in NewZealand by, among othem implicationsof availablehistorical and palcoseismic data for Peeket al. [!980]and Smith and Berryman [1986]. analysisof seismicityrates and fault behavior. Studiesof individualfaults have shown that therecurrence r• oflarge earthquakes (M>7) may be under-predicted bythe/• value SeismicityRate Analysis model[e.g., Wesnousky et al., 1983;Schwartz and Coppersmith, 1984;Davison and Scholz, 1985]. Rather, the frequency of these Thetraditional approach toseismicity rate analysis isto assume large events appears to be governedby relationsbetween f•ult a uniformspatial distribution of earthquakes and an exponential geometry, segmentation, andcumulative geological offset, which distributionof magnitudes ofthe Gutenberg-Richter form: influencethe distribution of strength properties along fault zo• COWANET AL.: PALEOSEISMICITYIN A NEWLYFORMED FAULT ZONE 6031

[e.g.,King and Nabelek, 1985; Sibson, 1989; Wesnousky, 1988, (~25 mm/yr) isequivalent tomore than 60% of the total rate of 1990].The recognition ofstructural controls onthe propagation, relative plate motion [DeMets etal., 1990], which represents the •est,and subsequent slipdistribution associated with some highest rate of slip in the Marlborough Fault System [Van Dissen c0seismicfaultruptures [e.g., Sibson, 1985, 1989] has drawn and Yeats, 1991]. The last surface rupture inzone 1occurred onthe attentiontothe significance of fault segmentation asa possible Hope River segment during an M -7.3earthquake in 1888 (Figure indicatorofthe size of earthquakesassociated with a givenfault. la) [McKay,1890; Cowan, 1991]. The Hope River segment has Thisconcept isimplicit in the model of"characteristic"earthquake rupturedrepeatedly at intervals of 80-200years during the last recurrenceinwhich most of the deformation associated with a given millenium,orapproximately every 140 years during the last 3500 faultor segment isreleased coseismically byearthquakes that occur years,if the sliprate and local offset in the 1888event are included withina relativelynarrow recurrence time and magnitude[Cowan and McGlone, 1991] (Table3). distribution[Working Group on CaliforniaEarthquake Zone 2 is centered on the PPAFZ whose cumulative offset and Probabilities,1990]. Other models advanced toexplain geological slip rateis approximatelyone order of magnitudesmaller than the evidenceof time-variable and slip-variable fault behavior [e.g., Hope fault. Both zone 1 and zone 2 containadditional faults Schwartz,1989; Scholz, 1989] also imply the nonrandomassociated with Holocene surface traces, for which litfie or no occurrenceof large earthquakes. palcoseismicdata are available[Wood et al., 1994; this study]. Thereis considerabledebate as to howwidely and over what Uncertaintiesin the locationof historicearthquakes mailer than timescalethe "characteristicearthquake" and other modelsof magnitudeM 4.0, precludethe selectionof smallerzones adjacent periodicfault behavior may apply [cf. Nishen'ko, 1991; Nishenko to the major faults [cf. Wesnousll.7,1994]. However, the depth andSykes, 1993; Kagan and Jackson, 1994, 1995; Reelefts and distributionof seismicityrecorded during microearthquake surveys Langbein,1994]. While thebehavior of somefaults appears to be in zone 1 [Arabaszand Robinson, 1976; Kieckhefer, 1977] andzone g0vemedby characteristicearthquakes over several earthquake 2 [Reynersand Cowan, !993] indicatesthat most seismicityis cycles[e.g., Schwartz and Coppersmith,1984] and othersby restrictedto the upperi0-15 km of the crust.By conservatively temporEclusters of earthquakes[e.g., Wallace, 1987; Coppersmith, choosing 50 km widezones for seismicityrate analysis in thisstudy, 1988;Sieh et aI., 1989], comparisonsof seismicityat the scaleof we are thereforeconfident that all seismicityassociated with the plateboundary segments have indicated nonperiodic fault behavior majorseismogenic faults has been sampled. [Kaganand Jackson, 1995]. Recentanalysis of historicseismicity and geological fault-slip SeismicityDistribution and RecurrenceParameters datafrom southern California [Wesnousky, 1994] indicatesthat the Three time periods were selected based on the relative structurallymature faults of the San Andreasfault zonedo exhibit completenessof the seismicitycatalogue: for eventsM >5 since characteristicbehavior, at leastover the time periods (102-104 1942; M >4 since1964 and M >3 since1988. Only those evevts with years)considered. In NorthCanterbury, one historic surface rupture crustaldepths (i.e., restricted and free depths shallower than 4 km) ofthe Hope River segmentof theHope fault(Figure la), together were sampled,so as to excludeevents located in the subdueted withpalcoseismic data and spatial variations in lateQuaternary slip Pacificplate 100 km beneaththe Hope fault region [Robinson, ratefrom the same segment, are also consistentwith the 1991].The cataloguewas also truncated at May 30, !994, to avoid characteristicearthquake model [Cowan, 1990, 1991; Cowanand sampling those aftershocksof the lune 1994 Arthur's Pass McGlone,1991]. Elsewhere in New Zealand,significant variations earthquakethat fall within the northwesterncomer of zone 2 infault behavior have been documented [Berryman and Beanland, (Figurelb). !991],and the problem there, as in mostparts of theworld, is one The instrumentalseismicity catalogue for zone 1 containsa ofreconciling geological and seismological data spanning different numberof eventsof magnitudeM 4.0-6.4. mainly in the east. •e periods.The seismicity catalogue for New Zealand spans only Several of theseevents are associatedwith two swarms,the first of thelast 150 yearsfor large earthquakes,and the instrumental which occurredsouth of the Hope fault on May 22, 1948, and cataloguespans a muchshorter period. This problem is intractablecomprised six events (Mz,= 5.9, 6.4, 6.2, 5.7, 5.7, and5.8) withinthe andacknowledged in seismicityrate analysisand hazard spaceof 2 hoursand 24 min [Eiby,1982]. The secondswarm (April estimation.However, the historic catalogue for North Canterbury is 22-28,1973) consisted of fourevents (Mz,-- 5.2, 4.5,4.4, and5.0) in sufficientlylong to illustratesignificant discrepancies between anarea northeast of theHope fault [Reyners, 1989]. Only the largest recurrencerates for large earthquakes extrapolated from the historic event of that swarm lies within the northeastern comer of zone 1. catalogueversus recurrence rates implied by geological data. Neither of the two swarms can be attributed to movement on known faults,but the Seismological Observatory File locations suggest that lrrequency-MagnitudeRelations for North theyare probably unrelated to the Hopefault. CanterburySeismicity Oneinteresting feature of thecatalogue for zone1 is theabsence of eventsof any magnitudein the westernpart of the regionand SeismicityZones fartherwest, where the 1888 and 1929 rapturesoccurred (Figure lb). This quiescencewas alsoapparent during an earlierstudy of For the purposeof comparinghistoric seismicity and microseismicity[Arabasx and Robinson,1976] andmay reflecta Pa!eoseismicdata,two 100 x50 krn zones (Figures la andlb) were significantreduction of stressin that regionfollowing the large selected,based on the contrasting structural maturity and geological historic earthquakes. • ratesof the known faults. Zone 1 encompassesan~35- tan-ling The seismicitycatalogue for zone2 (Figurelb) clearlyshows •easingbend in the Hope fault zone across which about half (~ 10- diminishingactivity south of the PPAFZ,but the strike of thatzone 14rnm/yr) ofthe total Hope fault slip rate is localized onthe Hope is remarkablywell-defined as noted by Reyners[1989], despite all !•versegment, with the remainder distributed across adjacent but one event (M •6.2) [Eiby, 1990], being smaller than M 5.5.A splays,notably theKakapo fault [Cowan, !990; Van Dissen and recentstudy ofmicroseismicity inNoah Canterbury [Cowan, 1992; l'eats,1991; Yang, 1991] (Figure la). The Hope fault has a Reynersand Cowan, 1993; H.A. Cowan etal., manuscript in cumulativeoffsetof~20 km [Freund, 197!], and its total slip rate preparation,1995] has also shown seismicity clustered along the 6032 COWANET AL.: PALEOSEIS•• IN A NEWLYFORMED FAULT ZONE

Table3. Magnitude-Frequency RelationsWithin the Hope Fault Region (Zone 1) and Porter's Pass-to-Amberley FaultZone (Zone 2)

Period Numberof RecurrenceParameters ReturnPeriods (years) Threshold (Number of Events , , Years) MagnitudeRegressed a SE b SE M6 s M7 s M> 7 r

HopeFault Region (Zone 1) 1942-1994 M 5 10 3.76 0.47 0.87a 0.08 29 214 80-200b (52.5) 3.64 .0.47 0.85 0.08 29 204 !964-1994 M 4 4 3.3! 0.22 0.87a 0.05 8! 602 60-300 e (30.5) 3.22 0.22 0.85 0.05 76 537 !988-1994 M 3 32 4.33 0.23 1.20a 0.06 - - (6.5) 3.14 0.43 0.87 0.12 120 891

PPAFZRegion (Zone 2)

1942-1994 M 5 6 1.83 0.75 0.60 0.13 59 234 1300- (52.5) 3.33 0.88 0.87 0.16 78 575 2000d 1964-1994 M 4 20 3.01 0.35 0.81 0.07 71 457 750-1000 e (30.5) 3.29 0.36 0.87 0.08 85 631 1988-1994 M 3 48 3.81 0.13 0.97a 0.04 102 955 1500-2000f (6.5) 3.43 0.17 0.87 0.05 62 457

Exceptas otherwise noted, b wasdetermined through an iterative procedure in whichonly a wasallowed to vary. Two sets of recurrence parametersare given for each of the three time periods in order to evaluate the sensitivity of a andb values. The standard errors (SE) indicate nosignificant differences in a for differeravalues of b.The superscript S and P denotehistorical return periods implied by seismicityand palcoseismicdata, respectively. a The a andb caseswere determinedsimultaneously by a linear least squaresfit. t:Cowan and McGIone [ 1991]. ½Calculated using slip rate of 10-25m/kyr and slip per event of 1.5-3.0m. Dataare from Cowan and McGlone [1991] and Van Dissen and Yeats [1991]. dFrom landslide and fault rupture data (this study). e Calculatedusing a sliprate of 3-4 m/kyrand slip per event of 3 m. Slip ratesfrom Berryman [ 1979] and Knuepfer [1992]. f Calculatedusing a slip rate of 3-4 m/kyr and slip per event of 6 m.Slip rates from Berryman [1979] and Knuepfer [199o_].

PPAFZ,with most events conf•ed to the upper 8-15 km of thecrust. Gutenberg-Richtermodel (Figure 7), but theclosest match between The recurrenceparameters a and b of the Gutenberg-Richterthe histories/andgeological data and the smallesterrors for the relationwere computed for thePPAFZ and Hope fault regions using historicalrecurrence parameters are obtained from the magnitu• ordinaryleast squares regression and an iterativeapproach to the frequencydistribution of M >3 earthquakessince 1988 (Table 3). selectionof b values(Table 3). Therecurrence parameters listed in Therecurrence estimates for largeearthquakes derived from longer Table 3, and illustratedfor the period 1964-1994in Figure7, periodsof historicalseismicity in zone2 are higherthan those indicatenegligible differences in activityrates between the selected impliedby thepalcoseismic data (Table 3). regions,despite an orderof magnitudedifference in slip rate and cumulativeoffset. By contrast,the palcoseismicdata from each regionimply significantdifferences in recurrenceintervals for large Discussion earthquakes(Table 3), yet neithercould be reliablypredicted from the historicseismicity catalogue. Furtherinterpretation of the historical seismicity data may be In zone1 thepresence of manyadditional faults with potential to unwarrantedgiven that the seismicitycatalogue is possiblytoo generatelarge earthquakes in theHope fault region emphasizes the shortto sustain arobust "productivity" analysis atmagnitudes less deceptivenessof contemporary seismicity. Moreover, the thanM 4.The historical data in the magnitude range M <6.0straddle discrepencybetween recurrence rates for largeearthquakes inferred aperiod of seismicquiescence in the Hope fault region, which may from historicaland geologicaldata would be accentuatedif the notbe representative of the longer-term seismicity rate. Similarly, swarmearthquakes mentioned above were to be excludedfrom the therelatively high rate of seismicity inthe PPAP-Z dur'mg thelast 30 analysis;i.e., the seismicityrate above magnitude 5 wouldbe even yearsmay reflect long-term temporal variations indeformation and lower in zone 1 during the historicalperiod. The situationis clusteringofearthquake occurrence, asfor example, inferred from differentin zone2, wherethe discrepencybetween historic and comparisonsoflow geological sliprate but higher geodetic smi• palcoseismicdata is in the oppositesense (Table 3). There, the andseismicity rates, in thereverse fault province of northwestera recurrencerate for large earthquakesis approximatedby the SouthIsland [H. Andersonet al., 1993;Bearart et al.,1994]; COWANET AL.:PALEOSEISMI•• IN A NEWLYFORMED FAULT ZONE 6033

lO 10'7 PPAFZ(Zone2) HopeFault(Zone1) 1964-1994 lO o -ii. 'i "','' Geologicaldata 10-2 - ii ! ',, 10.-3

Geologicaldata

3 4 5 6 7 8 3 4 5 6 7 8 Magnitude Magnitude Figure7. Comparisonof frequency-magnituderelations for theHope fault (Zone 1) andPPAFZ (Zone 2) (Figure lb) for the period1964-1994. The frequencyhistograms were compiledat classintervals of M 0.1. The solidline representsthe cumulative frequency, whereas the dashed straight line denotes the best fit ( b value0.87) for the magnituderange defined by verticalticks. The shadedboxes indicate the recurrencerate and possible magnitude for largeearthquakes inferred from geologicaldata. The magnitude-frequencydistribution for the Hope fault is in accordwith thecharacteristic earthquake model, whereas that of thePPAP-Z is approximatedby theGutenberg-Rich- ter model.Note in Table3, however,that the Gutenberg-Richterrelationship overpredicts the recurrence rate of M >7 earthquakesin the PPAFZ.

Temporalvariations in seismicityrate may profoundlyaffect the ---2500years is only sparselypreserved. Moreover, while the frequency-magnitudedistribution and apparent relationshipscombined evidence of faultingand landslidesbetween 2000 and betweenseismic productivityand fault slip rate, structural 2500 yearsB.P. is quitecompelling, this cannot be saidof the500- complexity,or maturity. For example, had it beenpossible to sample 700 yearB.P. event, for whichour interpretationrests largely on the seismicityin the Hopefault region for the50 yearsstraddling the landslidedata alone. 1888earthquake, the productivity could have been higher and the If ourrecord of lateHolocene rupture of thePPAFZ is complete, frequency-magnituderelationships implied by such data might not theimplied interval of 1300-2000years between the last two events haveunderpredicted therecurrence rate for the largest events on the wouldrequke 4-8 m displacementper event to be consistent with Hopefault to thesame degree as the current data set (Table 3). theinferred slip rate of 3-4 mm/yron thePorter's Pass fault. Conversely,lowseismicity rates may typify the behavior of the Displacementsperevent of thisorder have been documented for Hopefault, since it isclear that the M--, 7.3 event of one century ago numerousHolocene faults throughout New Zealand [Bcrr3,'man and hasnot been followed byongoing high rates ofactivity. Beanland,1991] and fi,rnad6ttir etal. [1995] have modeled upto 5 Returnperiods for large rare events are clearly sensitive toslight m of displacementforthe 1994 Arthur's Pass earthquake (Figure changesinrecurrence parameters when regressions areperformed la). Thusthe inferred return period, displacement perevent, and ondata dominated by smallerevents, or whenthe catalogue is sliprate need not be inconsistent. Moreover, it seems reasonable to heterogeneouswith respectto magnitude[e.g., Okal and conjecturethat larger displacements perevent could be expected l•omanowicz,1994;Kagan and Jackson, 1995]. It isperhaps ironic withinanewly formed fault zone composed ofnumerous short (<10 thereforethat the closestmatch betweenthe historicaland km) segmentsthat might accommodate more elastic strain than a geologicaldata from the PPAFZ should be obtained from only 6.5 maturefault which has undergone strain softening. The behavior of years(1988-1994) of seismicitydata (Table 3), giventhat thePPAFZ at its present stage of development could therefore be recurrenceestimates based on such a short time period would be moreconsistent with that of intra-platefault zones, i.e., assignedalow weighting inany seismic hazard assessment ofthat characterized bylow slip rate, more interseismic faulthealing region. processes,andlarger slip or stress drop per event [Kanamori and Theavailable palcoseismic datashow that even at this early Allen, 1986]. stageofdevelopment, elements ofthe PPAFZ are amenable to Althoughtheavailable recurrence anddisplacement estimates palcoseismicinvestigation. The apparentabsence of large for largeearthquakes in the PPAFZ are consistentwith the inferred landslidesorsurface ruptures with ages between oryounger than geologicalslip rate of thePorter's Pass fault, we cannot discount the 500-700and2000-2500 years B.P. indicates that the data for the late possibilitythat the slip per event is lessthan that required to account Holocenemay be complete.However, it is clearfrom the for the total strain, which recent studies have shown to be distributionof landslide ages that evidence for events older than geodeticallymeasurable at well over the 95% confidencelevel 6034 COWAN ET AL.: PALEOSRISMICITY IN A NEWLY FORMED FAULT ZONE

(Pearsonet al., submittedmanuscript, 1995). At present,slip rates Stephenson,Preliminary reports on the Inangahua earthquake, New for faults in the centraland easternparts of the PPAPZare too Zealand,May 1968,DSIR Bull. 193, 1968. .•rnaike,F., Seismicexplorations of the buried fault associated withthe poorly definedto permit momentrate summationfor direct 1948 Fukui earthquake,J'. Phys. Earth, 35, 285-308, 1987. comparisonwith rates of geodeticstrain, but it is possiblethat some Anderson,H., T. Webb,and J. Jackson, Focal mechanisms oflarge fraction of the slip across the PPAFZ is distributedamong quakesinthe South Island of New Zealand: Implications forthe aceore. secondarystructures, or accommodatedby smallerearthquakes modationof Pacific-Australiaplate motion, Geophys. J. Int., 115,1032. 1054, 1993. thanwe presently recognize in thegeological record. One of several Anderson,H., S.Beanland, G. Blick, D. Darby,G. Downes, I. Haines, J. explanationsadvanced to explain a historicdeficit of moment Jackson,R. Robinson,and T. Webb,The 1968 May 9_3 Inangahua, New releasein California[e.g., Dolan et al., 1995]is thata largerfraction Zealand,eaxthquake: An integratedgeological, geodetic, and seisra0. of the total sliprate across that region is releasedby morefrequent logicalsource model, N•Z. J. Geol. Geophys.,37, 59-86, 1994. moderate(M~6-6.5) earthquakesfor whichthere is lessgeological Pmderson,J.G., M. Ellis,and C. DePolo,Earthquake rate analysis, TeetonG. evidence [Ward, 1994; Wesnousky,1994]. Alternatively, a physics,218, 1-21, 1993. Arabasz,W.J., and R. Robinson,Microseismicity and geological strucl• proportionof the strain may be accommodatedby relatively inthe northern South Island, New Zealand, N. Z.J. Geol.Geophys., I9, aseismicslip in numeroussmall earthquakes[e.g., Lin and Stein, 569-601, 1976. 1989;Hill et at., 1990]. In thePPAFZ, the historicalseismicity rates ,&rnad6ttir,T.,J. Beavan, and C. Pearson, Deformation associated withthe may be "randomly"high for the sampledtime periods and 18 June,!994, Arthur'sPass earthquake, New Zealand,N.Z.J. Geol. Geopirys.,in press,1995. magnituderange considered,but it is also plausiblethat the Barnes,P.M., Structuralstyles and sedimentation at thesouthern termina- frequencyof largeearthquakes may be quasi-periodic and governed tionof theHikurangi subduction zone, offshore North Canterbury, byfactors unique to a fault zonein theearly stages of development. Zealand,Ph.D. thesis, Univ. of Canterbury,Christchurch, New Zealand, 1994. Conclusions Beavan,J., D. Darby,G. Blick,C. Pearson, T •d'nad6ttir, R.I. Walcott, Parsons,and P. England,GPS measurementsat the transitionbetween Dated surfaceruptures and landslidescorroborate an inference continentalcollision and subduction on thePacific-Australian plate boundaryinNew Zealand, Eos Trans. AGU, 75(44), Fall Meet. Suppl, of twolarge (M >7.0)earthquakes within the PPAFZ during the last 669, 1994. ~2500 years.Two earlierevents during the Holoceneare also Berryman,K., Activefaulting and derivedPHS directionsin theSouth recognized,but the dataprior to 2500 yearsare presumedto be Island,New Zealand, in TheOrigin of theSouthern Alps edited byR.[ Walcottand M.M. Cresswell,Bull. R. $oc.N.Z., 18, 29-34,1979. incomplete.A returnperiod of 1300-2000years for ruptureof the Berryman,K.R., andS. Beanland,Variation in faultbehavior in different PPAFZwould be consistentwith theinferred slip rate (3-4 rnm/yr) tectonicprovinces of New Zealand,J. Struct.Geol., 13, 177•189,1991. of itsprincipal element, the Porter's Pass fault, provided that the slip Brown,LJ., Geologicalmap of NewZealand, sheet S76 "Kaiapoi", 1sted. per eventis in the range4-8 m. scale1:63,360, Dep. of Sci.Ind. Res., Wellington, New Zealand, 1973. Palcoseismicdata from the adjacentstmcturally mature Hope Bull,W.B., J. King, E Kong,T Moutoux,and W.M. Phillips, Lichen dating of coseismiclandslide hazards in alpinemountains, Geomorphology, fault zone farthernorth indicatesignificant differences in fault 10, 253-264, 1994. b•haviorwith respectto the PPAF'Z,with recurrenceintervals for Burrows,CJ., A 500-yearold landslide in theAcheron River valley, Caw largeearthquakes apparently differing by a factorof 5 or more. terbury,N.Z. J. Geol.Geophys., 18, 357-360, 1975. Paradoxically,the differences in historicalseismicity rates for the Carter,R.M., and L. Carter,The MotunauFault and otherstructures at southernedge of the Australian-Pacificplate boundary, offshore Marl- Hopefault and PPAFZ ( zones i and 2, respectively, Figurelb) are borough,NewZealand, Tectonophysics, 88,133-159, 1982. negligibledespite an order of magnitudedifference inslip rate and Chinn,T.J.H., Use of rock weathering-rind thickness forHolocene absolu!e cumulativeoffset associated with the major faults. The magnitude- age-datingin NewZealand, Arct. Alp. Res., 13, 33-45,1981. frequencydistribution for theHope fault region is in accordwith the Coppersmith,K.I., Temporaland spatial clustering of earthquakeactivity the centraland eastern United States, Seismol. Res. Lett., 59, 299-304, characteristicearthquake model, whereas the magnitude-frequency 1988. distributionin the PPAPZ is approximated,but slightly Cornell,C.A., Engineeringseismic risk analysis, Bull. Seismol. Soc. overpredicted,by theGutenberg-Richter relationship (Table 3 and 58, 1583-1606, 1968. Figure7). The comparisonof thesetwo fault zonesdemonstrates Cowan,H.A., LateQuaternary displacements onthe Hope fault at Gly•m theimportance of thestructural maturity of thefault zone in relation Wye,North Canterbury, N.Z.J. Geol.Geophys., 33, 285-293,1990. to seismicityrates inferred from recent, historical, and palcoseismic Cowan, H.A., TheNorth Canterbury earthquake of September1, 1888, J. R. Soc.N.Z.,21, !-12, 199!. Cowan,H.A., Structure,seismicity and tectonicsof the Porter's AmberleyFault Zone, North Canterbury,New Zealand.Ph.D. th•s, Acknowledgments.This study was fundedby the New Zealand Univ. of Canterbury,Christchurch, New Zealand,1992. UniversityGrants Committee and New Zealand'Earthquake Commission, Cowan,H.A., (Compilor),Field guideto New Zealandactive tecto•es- andformed part of a Ph.D. programmewithin the Active Tectonics Research IASPEI94, 27thGeneral Assembly of IASPEI,January 1994, Welllag. Groupat the Universityof Canterbury.Brian Molloy andNeville Moar ton, R. Soc.N.Z. Misc. Sen, 27, 1994. providedradiocarbon ages and pollendata from sitesat Porter'sPass. We Cowan,H.A., andM.S. MeGlone,Late Holocene displacements and thankWarwick Smith and the New Zealand Seismological Observatory for acteristicearthquakes on the Hope River segmentof theHope fault, seismicitydata and Conrad Lindholm for helpwith datareprocessing at New Zealand,J.R. Soc.N.Z., 21,373-384, 1991. NORSAR.Reviews by JohnAdams, Kelvin Berryman, Bill Bull,Hilmar Coyle, S., The Porter'sPass Fault, M.Sc. thesis,Univ. of Canterbury, Bungum,JargPettinga, Maud McSaveney, MarkStirling, andBob Yeats Christchurch,NewZealand, 1988. improvedthepaper and provided impetus toextend some aspects ofthe Davis,T.L., and S.S. Namson, A balanced cross-section ofthe I994 work,which was completed whilethe first author worked atthe Norwegian Northridge earthquake, southern California, Nature, $72, 167-169, GeotechnicalInstitute and NorwegianSeismic Array (NORSAR), 1994. supportedby a fellowshipfrom the Research Council of Norway. Davison,F.C., and C.H. Scholz,Frequency-moment distribution of quakesin the AleutianAre: A test of the characteristicearthq• References model,Bull. Seisrnol.Soc. Am., 75, 1349-1361,1985. DeMets,C., R.G. Gordon, D.E Argus,and S. Stein,Current plate moti0.• Adams,J., ,--Earthquake-dammedlakesin NewZealand, Geology, 9, 215- Geophys.J. !nt., I01,425-478, 1990. 219, 1981. Doig,R., 2300yr historyof seismicityfrom silting events in Lake Tadot• Adams,R.D., G.A. Eiby, M.A. Lowry,G.J. Lensen, R.P. Suggate,and W.P. sac,Charlevoix, Quebec, Geology, 18, 820-89_.3,1990. COWAN ET AL.: PALEOSEISMICITY IN A NEWLY FORMED FAULT ZONE 6035

Dolan,J.F., K. Sieh,T.K. Rockwell, R.S. Yeats, J. Shaw,J. Suppe,G.H. NewZealand Seismological Observatory, New Zealand earthquakes: 1850- I/u/tile,and E.M. Garb,Prospects for largeror more frequentearth- 1994.Observatory Master File, Inst. of Geol.and Nuel. Sci., Welling- quakesinthe Los Angeles metropolitan region, Science, 267, 199-205, ton, New Zealand, 1994. 1995. Nicol,A., B.V.Alloway, and P.J. Tonkin, Rates of deformation,uplift and Dowrick,D.J., and E.G.C. Smith, Surface wave magnitudes of some New landscapedevelopment associated with activefolding in theWaipara Zealandearthquakes 1901-1988, Bull. N. Z. Nat.Soc. Earthquake Eng., areaof NorthCanterbury, New Zealand,Tectonics, 13, 1327-1344, 23,198-219,1990. 1994. Eiby,G.A., An annotatedlist of NewZealand earthquakes, 1460-1965, Nishenko, S.P., Circum-Pacifi½ seismic potential: 1989-1999, Pure AppI. N•.J. Geol.Geoœhys., 11,630-64'7, 1968. Geoptrys.,135,169-259, 1991. Eiby,G.A., New Zealand seismological report 1948-49~50, Bull. E-164, Nishenko,S.P., and L.R. Sykes, Comment on "Seismic gap hypothesis: Ten S•ismol.Ohs., W½11ington, New Zealand,1982. yearsafter" by Y.Y. Kaganand D.D. Jackson,J. Geophys.Res., 98, Eiby,G.A., The Lake Coleridge earthquakes of 1946, Bull. N.Z. Nat. $oc. 9909-9916, 1993. EarthquakeEng., 23, 150-158,1990. Okal,E.A., andB.A. Romanowicz,On thevariation of b-valueswith earth- Fr•und,R., TheHope Fault: A strike-slipfault in NewZealand, N.Z. Geol. quakesize, Phys. Earth Planet. Inter., 87, 55-76,1994. Surv.Bull., 86, 1971. Pearson,G.W., and M. Stuiver,High-precision calibration of theradiocar- 6arlick,R., LeesValley Fault. B.Sc.(Hons)dissertation, Univ. of Canter- bontime scale, 500-2500 BC, Radiocarbon, 28,839-862, 1986. bury,Christchurch, New Zealand, ! 992. Peek, R., J.B. Benill, and R.O. Davis, A seismicitymodel iiar New Crregg,D.R., Geological map of NewZealand sheet 18, "Hurunui" scale Zealand,Bull. N.Z. Nat.$oc. Earthquake Eng., !3, 355-364,1980. 1:250,000,N.Z. Dep. of Sci. Ind. Res.,Wellington, New Zealand,1964. Plafker, G., and G.E. Ericksen, Nevados Huascaran avalanches, Peru, in Henderson,J., West Nelson earthquake of 1929,N.Z.J. $ci. Technol.,19. Rockslidesand Avalanches1, editedby B. Voight,pp. 277-314, Elsevier 66-144, 1937. Sci., New York, 1978. ttcu'zer,R.H., andJ.D. Bradshaw,The MotunauFault and other structures at Reyners,M.E., New Zealandseismicity 1964-87: An interpretation,N.Z.J. the southernedge of the Australian-Pacificplate boundary,offshore Geol. Geophys.,32, 307-315, 1989. Marlborough,New Zealand,Tectonophysics, J15, 161-166,1985. Reyners,M.E. and H.A. Cowan,The transitionfrom subductionto conti- Hill,D.P., J.P. Eaton, and L.M. Jones,Seismicity !980-86, in The San nental collision:crustal structure in the North Canterburyregion, New AndreasFault System,California, U.S. Geol. Surv.Prof. Pap. 1515, Zealand,Geopl•ys. J. Int., 115, 119_4-1136,1993. 115-15!, 1990. Robinson,R., Extent andgeometry of subductionin the northernSouth Hull,A.G., Tectonicsof the 1931 Hawk½'sBay earthquake,N.Z.g. Geol. Islandand Wellingtonregions, New Zealand,Geol. Soc. N7.. Misc. Geophys.,33,309-320, 1990. PubI., 56, 44-46, 1991. Jibson,R.W., C.S. Prentice,B.A. Borisoff,E.A. Rogozhin,and C.J.Langer, Roelofts, E., and J. Langbein,The earthquakeprediction experiment at Someobservations of landslidestriggered by the 29 April, 1991, Racha Parkfie!d,California, Rev. Geophys., 32, 315-336, 1994. earthquake,Republic of Georgia,Bull. Seismol.Soc. Am.. 84,963-973, Rynn, J.M.W., and C.H. Scholz, Seismotectonicsof the Arthur's Pass I994. region,south Island, New Zealand,Geol. Soc. Am. Bull., 89, 1373-1388, Kagan,Y.Y., and D.D. Jackson,Long-term probabilistic forecasting of 1978. earthquakes,J. Geophys. Res., 99, 13,685-13,700,1994. Scholz,C.H., Commentson models of earthquakerecurrence, in, Proceed- Kagan,Y.Y., and D.D. Jackson,New seismicgap hypothesis: Five years ingsof ConferenceXLV- Fault Segmentationand Controlsof Rupture after,J. Geophys.Res., J 00, 3943-3959,1995. Initiation and Terrnination,edited by D.P. Schwartzand R.H. Sibson, Kanamori,H., and C.R. Allen, Earthquakerepeat time and averagestress UJ. Geol.Surv. Open File Rep.,89-315, 350-360, I989. drop,in EarttuluakeSource Mecimnics, Geophys. Monogr. Set., vol 37, Schwartz, D.P., Palcoseismicity,persistence of segments,and temporal editedby S. Das, J. Boawright,and C.H. Scholz,pp. 157-167,AGU, clusteringof large earthquakes-Examplesfrom the San Andreas, Washington,D.C., 1986. Wasatch,and Lost River fault zones,in, Proceedingsof ConferenceXL,V Keefer,D.K., Landsfidescaused by earthquakes,Geol. $oc. Am. Bull., 95, - Fault Segmentationand Controlsof RuptureInitiation and Termina- 406-421, 1984. tion, editedby D.P. Schwartzand R.H. Sibson,U3'. Geol. Surv.Open Kieckhefer,R.M., Microseismicityin the vicinity of the Clarencefault, File Rep.,89-315,361-375, 1989. NewZealand, N.Z. J. Geol. Geophys.,20, 165-177,1977. Schwartz,D.P., and K.J. Coppersmith,Fault behaviorand characteristic King,G.C.P., The accommodation of large strains in theupper lithosphere earthquakes:Examples from Wasatch and San Andreas Faults, J. Geo- ofthe earth and other solids by self-similarfault systems: The geometri- phys.Res., 89, 5681-5698,1984. calorigin of b-value,Pure AppI. Geophys., 121,762-815, 1983. Shaw,J., andJ. Suppe,Active faultingand growthfolding in the eastern King,G.C.P., and J. Nabelek,The roleof bendsin faultsin theinitiation SantaBarbara Channel, California, Geol. Soc.Am. Bull., 106, 607-626, andtermination of earthquakerupture, Science 228, 984-987, 1985. 1994. Kirkaldy,P.H.S., and E.G. Thomas, Final report on a seismicsurvey in the Sibson,R.H., Stoppingof earthquakeruptures at dilationalfault jogs, CanterburyPlains area of NewZealand, BP Shell and Todd Petroleum Nature, 316, 248-251, 1985. DevelopmentLtd, N.Z. Geol. Surv. Open File Pet. Rep., 328, 1963. Sibson,R.H., Earthquakefaulting as a structuralprocess, J. Struct.Geol., Knuepfer,P.L.K., Temporal variations in latest Quaternary slip across the !I, 1-14, 1989. Australian-Pacificplateboundary, northeastern South Island, New Sieh,K., M. Stuiver,and D. Brillinger,A more precisechronology of earth- Zealand,Tectonics, 11,449-464, 1992. quakesproduced by the SanAndreas fault in southernCalifornia, J. Lin,J., and R.S. Stein,Coseismie folding, earthquake recurrence and the Geophys.Res., 94, 603-623,1989. 1987source mechanism at WhittierNarrows, Los Angeles Basin, Cali- Skey,H.E, Reporton theNorth Canterbury earthquake of 25thDecember, 199,2,in RecordSurvey N.Z., Dept. of Landsand Surv., Wellington, f0mia,J. Geophys.Res., 94, 9614-9632, 1989. New Zealand, 1925. McKay,A.,On the earthquakes ofSeptember 1888, in the Amuri and Mar- Sinalley,R., J. Pujol,M. Regnier,J.M. Chiu,J.L. Chatelain,B.L. Isacks,M. lboroughdistricts of theSouth Island, N•'. Geol.Surv. Rep. Geol. Araujo,and N. Puebla,Basement seismicity beneath the AndeanPre- ExpIor.,20, 1-16, 1890. cordilleranthin-s 'kinned thrust belt and implicationsfor crustal and McKay,A., Reporton the recent seismic disturbances within Cheviot lithosphericbehavior, Tectonics, 12, 63-76,!993. Countyinnorthern Canterbury and the Amud District of Nelson, New Smith,W.D., and K.R. Berryman,Earthquake hazard in New Zealand: Zealand,Govt. Print., Wellington, New Zealand, 1902. lnlErencesfrom seismology and geology, Bull. Roy. Soc. N.Z., 24,223- MeSaveney,M.J., Sherman Glacier rock avalanche, Alaska, U.S.A,, in 242, 1986. RockslidesandAvalanches 1, edited by B.Voight, pp. 197-258, Elsevier Speight,R., TheArthur's Pass earthquake of 9thMarch, 1929, N.Z.J. Sci. Sei.,New York,1978. Technol.,15, 173-182, 1933. M½Saveney,MJ.,A manualof weathering-rind dating for sandstone ofthe Speight,R., Recentfaulting in theSouthern Alps, N.Z.J. Sci.TechnoI., !9, TorlesseSupergroup, Tech.Rep. 92/4, N.Z. Inst. of Geol. and Nucl. Sci., 701-708, 1938. Wellington,New Zealand, 1992. McSaveney,M.J.,Rock avalanches of 2 Mayand 16 September 1992, Stein, R.S., and G. EkstrOm,Seismicity and geometryof a 110 ima-long MountFleteher, New Zealand, Landslide News, 7, 2-4,1993. blind thrustfatdt, 2, Synthesisof the 1982-1985California earthquake MeSaveney,M.J., T.J. Chinn, and G.T. Hahcox, Mount Cook rock ava- sequence,J. Geophys.Res., 97, 4865-4883,1992. lancheof 14December 1991, New Zealand, Landslide News, 6, 32-34, Stein,R.S., and R.S. Yeats,Hidden earthquakes, Sci. Am., 260 (6), 30-39, 1989. 1992. 6036 COWAN ET AL.: PALEOSEISMICITYIN A NEWLY FORMED FAULT ZONE

Stuiver,M., and G.W. Pearson,High-precision calibration of the radiocar- W'dson,D.D., Geology ofthe Waipara subdivision, N.Z.Geol. Surv. Bull. bon time scaleAD 1950-500 BC, Radiocarbon,28, 805-838, 1986. 64, 122 pp., 1963. Sruiver,M., andP.J'. Reimer, A computerprogram for radiocarbonage cali- Wood,R.A., J.R. Pettinga, S.Bannister, G.Lamarche, andT.$. McMorraa, bration, Radiocarbon, 28, 1022-1030, 1986. Structureofthe Hahruer strike-slip basin, Hope fattit, New Zealand, Tortkin,P.S., and L.R. Basher,Soil-stratigraphic techniques in the studyof Geol. $oc. Am. Bull., 106, 1459-1473, 1994. soil and landformevolution across the SouthernAlps, New Zealand, WorkingGroup on California Earthquake Probabilities, Probabilities of Geomorphology,3, 547-575, 1990. largeearthquakes in the San Fransisco Bay Region, California, Trangrnar,B.B., Overviewof Canterburyloess deposits: Tour guide for gut. Geol. $urv. Circ., 1053, 51 pp., 1990. Syrup.on Loess:February 14-21, presented at Third Meetingof West- Yang,J.S., The Kakapo Fault-A major active dextral fault in the central ern Pacific, INQUA Loess Comm., Christchurch,New Zealand, 1987. NorthCanterbury-Buffer regions of NewZealand, N.Z.J. Geol. VanDissen, R., andR.S. Yeats,Hope Fault, Jordan thrust, and uplift of the phys.,34, 137-143, 1991. SeawardKaikoura Range, New Zealand,Geology, 19, 393-396,1991. Yang,J.S., Landslide mapping andmajor earthquakes onthe Kakapo Wallace,R.E., Groupingand migration of surfacefaulting and variations in SouthIsland, New Zealand, J. R.. Soc.N.Z., 22,205-212,199o_ sliprates on faultsin the GreatBasin province, Bull. $eismol.$oc. Am., Yeats,R.S., mid K.R. Berryman, South Island, New Zealand, and Trans. 77, 868-876, 1987. verseRanges, California: A seismotectonic comparison, Tectonics, 6, Whrd,S.N., A multidisciplinaryapproach to seismichazard in southern 363-376, 1987. California,Bull. $eismol.$oc. Am., 84, 1293-1309, 1994. Youngs,R.R., and K.J. Coppersmith, Implications of fault sllp rates Wellman,H.W., Datafor the studyof Recentand late Pleistocene faulting earthquakerecurrence models toprobabilistic seismic hazard estimale• in the southIsland of New Zealand,N.Z.J. $ci. Technol.B 34, 270-288, Bull. Seismol.5oc. Am., 75, 939-964, 1985. 1953. Yousir,H.S., The applications ofremote sensing togeomorphologieal he0. Wells,D.L., andK.J. Coppersmith,New empiricalrelationships among tectonicmapping in NorthCanterbury, New Zealand, Ph.D. thesis, magnitude,rupture length, rupture width, rupture area, and surface dis- Univ. Canterbury,Christchurch, New Zealand,1987. placement,Bull. $eismol. $oc. Am., 84, 974-1002, 1994. Wesnousky,S.G., Seismologicaland structuralevolution of strike-slip faults,Nature, 335, 340-343, 1988. H. Cowan,Norwegian Seismic Array, RO. Box 51, Kjeller,N-2007 Wesnousky,S.G., Seismicity as a functionof cumulativegeologic offset: Norway. (e-mail: [email protected]. no) someobservations from southernCalifornia, Bull. Seismol.$oc. Am., A. Nicol, Fault AnalysisGroup, Department of EarthSciences, Univer- 80, !374-1381, 1990. sity of Liverpool,Liverpool L69 3BX, England. Wesnousky,S.G., The Gutenburg-Riehteror characterisle earthquake dis- P. Tonkin,Department of Soil Science,Lincoln University, P.O. Box 94, tribution,which is it? Bull. Seismol.Soc. Arn., 84, 1940-1959,1994. Lincoln, New Zealand. Wesnousky,S.G., C.H. Seholz, K. Shimazaki,and T. Matsuda,Earthquake frequencydistribution and the mechanics of faulting,J. Geophys.Res., 88, 9331-9340, 1983. Whitehouse,I.E., andG.A. Griffiths,Frequency and hazard of largerock avalanchesin the centralSouthern Alps, New Zealand,Geology, 11, (ReceivedSeptember 23, 1994;revised May 15, 1995; 331-334, !983. acceptedMay 23, 1995. )